henhaobio
Sign in

Developmental ScienceNeurosciencePeer-reviewed journal article

Visual Cortical Response Variability in Infants at High Familial Likelihood for Autism

Abigail H. DickinsoniD, Madison BoothiD, Scott HubertyiD, Declan Ryan, Alana CampbelliD, Jessica B. GiraultiD, Neely C. MilleriD, Bonnie K. LauiD, John Martin ZempeliD, Sara Jane WebbiD, Jed T. ElisoniD, Adrian K. C. LeeiD, Annette M. Estes, Stephen R. DageriD, Heather C. HazlettiD, Jason J. WolffiD, Robert Schultz, Natasha M. MarrusiD, Alan Charles EvansiD, Joseph PiveniD, John R. PruettiD, Shafali JesteiD, IBIS Network

PDFDOIPreprint
In-depth full-text digest available · Jump to digest ↓

Abstract

Visual processing undergoes rapid development in the first year of life, supporting the emergence of higher-order cognitive, language, and motor functions. Visual evoked potentials (VEPs) provide a non-invasive measure of visual system maturation that may shed light on heterogeneous developmental trajectories among infants at high familial likelihood for autism. Infants with an older sibling with autism spectrum disorder (N = 177 at 6 months; N = 132 at 12 months) participated in the Infant Brain Imaging Study-Early Prediction (IBIS-EP) study. Pattern-reversal VEPs were recorded at 6 and 12 months, and developmental skills were assessed at 24 months using the Bayley Scales of Infant and Toddler Development (Bayley-4). VEP components (P1 and N1) were characterized by their amplitude and latency, as well as trial-to-trial variability in these measures. Associations with 24-month cognitive, language, and motor scores were examined using general linear models controlling for age, site, sex, and trial count. Robust VEPs were observed at both time points, with age-appropriate morphology and expected developmental changes, including decreases in P1 latency and amplitude from 6 to 12 months. Greater trial-to-trial variability in P1 latency at both time points was associated with higher cognitive and language scores at 24 months. In contrast, conventional measures of mean P1 latency and amplitude were not associated with developmental outcomes. These findings suggest that temporal variability in early visual responses may index adaptive sensory-circuit flexibility during a period of rapid experience-dependent development. VEP response-timing variability may therefore provide an early mechanistic marker of sensory-circuit organization relevant to later developmental trajectories. Research HighlightsO_LIGreater trial-to-trial variability in visual cortical response timing was associated with higher cognitive and language scores in infants at high familial likelihood for autism. C_LIO_LIConventional average VEP measures were unrelated to outcomes, suggesting temporal variability may more sensitively capture relevant neural-circuit differences in this population. C_LIO_LIGreater variability in early visual responses may reflect adaptive sensory-circuit flexibility during a critical period of neurodevelopment. C_LIO_LIVEP response-timing measures provide an early mechanistic window into sensory-circuit organization and its relation to later developmental trajectories. C_LI

In-depth summary

Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 2 (21 Jul 2026) · CC BY 4.0

Among infant siblings of autistic children, greater trial-to-trial variability in visual P1 latency at 6 and 12 months tracked higher 24-month cognitive and language scores; mean P1 measures did not.

EEG visual evoked potentials (VEPs) have been linked to later development in infants at elevated likelihood for other biological or environmental reasons, and resting alpha power to language in infants with an autistic older sibling, but whether evoked visual responses relate to outcomes in these high familial likelihood (HL) infants was unclear. It tests this in a five-site prospective cohort, adding single-trial timing variability to averaged P1 amplitude and latency.

Key results

  1. ShownInfants whose P1 latency jittered more from trial to trial, at both 6 and 12 months, had higher Bayley-4 cognitive and language composites at 24 months. The effect held with the other three P1 metrics in the same model.

    6-month EEG (98 infants): cognitive b = 7.34, p < .001, q = .005; language b = 8.81, p = .005, q = .028. 12-month EEG (97 infants): cognitive b = 4.54, p = .008, q = .050; language b = 6.95, p = .007, q = .050. Predictors standardized; models also adjusted for age at EEG, site, sex and usable trial count. · 3.2. Associations between VEP metrics and…; 2.5. Statistical Analyses; 2.1. Participants

  2. ShownAveraged P1 latency, averaged P1 amplitude and P1 amplitude variability showed no FDR-surviving association with any 24-month outcome at either age. The link between 6-month P1 latency variability and motor scores was only nominal.

    All q ≥ .092 for mean P1 latency, mean P1 amplitude and P1 amplitude variability across cognitive, language and motor outcomes. 6-month P1 latency variability vs motor: b = 3.55, p = .038, q = .092. · 3.2. Associations between VEP metrics and…; 4. Discussion

  3. ShownFrom 6 to 12 months, mean P1 latency, mean P1 amplitude and P1 amplitude variability fell, while P1 latency variability rose.

    Linear mixed models with site, sex and trial count as covariates (177 and 132 infants with usable VEPs). Latency 109.42 to 105.41 ms, b = −4.22, p < .001. Amplitude 13.06 to 10.33 µV, b = −2.56, p < .001. Latency variability 15.59 to 16.89 ms, b = 0.87, p = .008. Amplitude variability 15.43 to 14.52 µV, b = −1.10, p = .015. · 3.1. Developmental Changes in P1 Metrics; 2.5. Statistical Analyses; 2.1. Participants

  4. SuggestedSecondary N1 analyses matched only partly and more weakly. At 6 months, N1 latency variability related positively to motor scores after correction, but its cognitive and language links did not survive. No 12-month N1 association survived.

    6-month N1 latency variability: motor b = 3.66, p = .007, q = .041; cognitive b = 3.54, p = .029, q = .115; language b = 4.84, p = .052, q = .125. 6-month N1 amplitude variability vs motor: b = −3.96, p = .002, q = .025 (negative). No 12-month N1 metric survived FDR correction. · 3.2. Associations between VEP metrics and…; 2.5. Statistical Analyses

Methods, data and reproducibility

Checkerboard pattern-reversal VEPs were recorded with high-density EEG at 6 and 12 months. P1 and N1 amplitude, latency and their trial-to-trial median absolute deviation (MAD) were taken at O1, Oz and O2. Mixed models tested age change. General linear models with all four P1 metrics plus covariates tested links to 24-month Bayley-4 scores, with Benjamini-Hochberg FDR correction per age.

  • IBIS-EP (Infant Brain Imaging Study-Early Prediction) infant siblings of children with autism: 6-month pattern-reversal VEP recordings plus 24-month Bayley-4 composites · human · n = 180 recorded, 177 usable, 98 with 24-month behavioral data
  • IBIS-EP infant siblings of children with autism: 12-month pattern-reversal VEP recordings plus 24-month Bayley-4 composites · human · n = 135 recorded, 132 usable, 97 with 24-month behavioral data
Methods & tools
EEGLAB, MATLAB, R, E-Prime (Psychology Software Tools), Net Amps amplifiers, HydroCel Geodesic Sensor Nets (EGI), Bayley Scales of Infant and Toddler Development, Fourth Edition (Bayley-4), Social Communication Questionnaire (SCQ), Autism Diagnostic Interview-Revised (ADI-R)
Reproducibility gaps
Processing used custom MATLAB scripts, and no code link is given. The text has no data-availability statement or repository and gives no software versions. Results are an interim cut of data available as of January 2025.
Limitations (5)
  • Only infants at high familial likelihood were analysed. Without a general-population comparison group, it is unknown whether the associations generalize or are specific to familial likelihood. · stated by the authors
  • Only continuous developmental scores were analysed. Links to categorical autism diagnosis are deferred until the full, adequately powered sample is available. · stated by the authors
  • White families were overrepresented, and parents tended toward high education and income. Because sociodemographic factors can shape both early development and how it is assessed, generalizability is limited and replication in more diverse cohorts is needed. · stated by the authors
  • This is an interim analysis of one ongoing cohort, with 98 and 97 infants in the outcome models and no independent replication sample. Both 12-month P1 associations sit exactly at the q = .050 threshold. · noted in this summary
  • The idea that variability reflects plasticity is not tested here. Latency MAD comes from single trials kept by a morphology filter after visual inspection, with 15-98 usable trials per infant at 6 months. Single-trial peak-picking noise is therefore not separated out; trial count was handled only as a covariate. · noted in this summary

Read this if you study infant EEG or early markers of autism and developmental delay and want evidence that single-trial timing variability in VEPs carries outcome information that averaged waveforms miss.

Citations
2
FWCI
8.26
Field percentile
top 2%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISOtten '24Hardiansyah '26Knight '24Beker '25Naik '23Godel '25
71% semantic matchNeuroscience
Katharina Otten, James Christopher Edgar, Heather L. Green+9 more·Developmental Cognitive Neuroscience·Dec 24, 2024· 4 citations